RESEARCH Open Access

Multi-locus sequence analysis of Mycoplasmacapricolum subsp. capripneumoniae for themolecular epidemiology of contagious caprinepleuropneumoniaLucía Manso-Silván1*, Virginie Dupuy1, Yuefeng Chu2 and François Thiaucourt1

Abstract

Mycoplasma capricolum subsp. capripneumoniae (Mccp) is the causative agent of contagious caprinepleuropneumonia (CCPP), a devastating disease of domestic goats. The exact distribution of CCPP is not knownbut it is present in Africa and the Middle East and represents a significant threat to many disease-free areasincluding Europe. Furthermore, CCPP has been recently identified in Tajikistan and China. A typing method with animproved resolution based on Multi-Locus Sequence Analysis (MLSA) has been developed to trace new epidemicsand to elucidate whether the recently identified cases in continental Asia were due to recent importation of Mccp.The H2 locus, a polymorphic region already in use as a molecular marker for Mccp evolution, was complementedwith seven new loci selected according to the analysis of polymorphisms observed among the genome sequencesof three Mccp strains. A total of 25 strains, including the two new strains from Asia, were analysed by MLSAresulting in the discrimination of 15 sequence types based on 53 polymorphic positions. A distance tree inferredfrom the concatenated sequences of the eight selected loci revealed two evolutionary lineages comprising fivegroups, which showed good correlation with geographic origins. The presence of a distinct Asian cluster stronglyindicates that CCPP was not recently imported to continental Asia. It is more likely that the disease has beenendemic in the area for a long time, as supported by historical clinical descriptions. In conclusion, this MLSAstrategy constitutes a highly discriminative tool for the molecular epidemiology of CCPP.

IntroductionContagious Caprine Pleuropneumonia (CCPP) is adevastating disease of goats included in the list of notifi-able diseases of the World Organisation for AnimalHealth (OIE). The first description of the disease datesback to 1873, in Algeria [1]. However, the etiologicagent, Mycoplasma capricolum subsp. capripneumoniae(Mccp) was only isolated and characterised a centurylater, in 1976 [2]. This may be explained by several fac-tors. Mccp is one of the most fastidious mycoplasmas togrow in vitro and cultures are often overgrown by con-comitant bacteria, hampering its isolation. Mccp mayalso be difficult to identify because it belongs to theMycoplasma mycoides cluster, a group of five closely

related mycoplasmas that are pathogenic for ruminants,also comprising M. mycoides subsp. mycoides “SmallColony”, M. mycoides subsp. capri (Mmc), M. caprico-lum subsp. capricolum (Mcc), and M. leachii. Theseorganisms share many genotypic and phenotypic traits,which can lead to problems for identification, particu-larly when applying classical diagnostic techniques.Furthermore, there has been disagreement over theaetiology of CCPP, which was attributed to Mmc for along time, generating great confusion. Therefore, muchattention must be given to historical clinical descriptionsin order to distinguish true CCPP from pleuropneumo-nia caused by other members of the M. mycoides clus-ter, often associated to other pathologies.All of this may also explain why the exact distribution

of CCPP is not known. What is certain is that the dis-ease is present in Africa and the Middle East, as

* Correspondence: lucia.manso-silvan@cirad.fr1CIRAD, UMR CMAEE, F-34398 Montpellier, FranceFull list of author information is available at the end of the article

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86 VETERINARY RESEARCH

© 2011 Manso-Silván et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

demonstrated by isolation (which remains the confirma-tory test required by the OIE) and molecular characteri-sation of Mccp strains. Although Mccp was shown to bea rather homogeneous taxon [3-6], two molecular mar-kers revealed some degree of heterogeneity amongstrains allowing the first studies on the molecular epide-miology of CCPP. Mccp strains showed a high degree ofpolymorphism between the 16S rDNA genes of theirtwo rrn operons, as compared to other members of theM. mycoides cluster, which was exploited to analyse themolecular evolution of Mccp [7]. A subsequent study onthe molecular epidemiology of CCPP was conducted bythis group based on the analysis of the H2 locus, whichallowed the discrimination of four groups that showed agood correlation with geographic origins [8]. Since thepublication of these studies, very few new strains havebeen made available, although some of them were ofparticular interest. First, the isolation of Mccp in theThrace region of Turkey in 2004 showed that there is arisk of introduction of the disease in Europe [9], and therecent outbreak of CCPP in Mauritius [10] confirmedthat Mccp is spreading, threatening many disease-freeareas. Furthermore, the presence of CCPP in east Asiahas recently been confirmed using molecular techniquesin Pakistan [11] and in Tajikistan [12] and several Chi-nese strains have been characterised as Mccp [13]. Inaddition, CCPP was recently confirmed in wild rumi-nants kept in a wildlife reserve in Qatar [14], question-ing the long believed strict host specificity of Mccp tothe domestic goat. All these recent findings have openednew interesting questions that could not be solved by16S rDNA or H2 locus analysis due to the limited dis-criminatory power of these molecular markers. A typingmethod with an improved resolution is required toaddress all the epidemiological questions that arise, aswell as to trace new epidemics.The first objective of this study was to develop a dis-

criminative tool for the molecular epidemiology ofCCPP. A method based on the analysis of severalgenetic markers that is known as Multi-Locus SequenceAnalysis (MLSA) was chosen for this purpose. The ori-gin of Mccp in east Asia was then investigated using theimproved MLSA to determine whether the recentlyidentified Mccp strains were the result of a recentimportation or were endemic to the region.

Materials and methodsMccp strains, culture conditions and sample preparationThe 27 strains analysed in this study (Table 1) werecharacterised as Mccp by specific PCR amplification[15]. Most of them had already been analysed in pre-vious studies [7,8,16] and corresponding 16S rDNA andH2 types are presented whenever available (Table 1).Mycoplasma strains were cultured in modified Hayflick’s

broth [6] at 37°C, 5% CO2 and were harvested in thelate exponential phase of growth. DNA was extractedfrom 3 mL culture or pleural fluid using the DNeasyblood and tissue kit (Qiagen GmbH, Hilden, Germany)and was diluted to a concentration of 1 ng/μL in sterile,deionised water for use as PCR templates.

Genome sequencing and identification of loci for MLSATo identify the new loci for MLSA three near-completegenome sequences of differing Mccp strains (9231-Abomsa, 95043 and 97095-Tigray) were obtained byGATC Biotech AG (Konstanz, Germany). The genomesequence of strain 9231-Abomsa was obtained by 454(Life Sciences, Roche, Basel, Switzerland) and verified bySolexa/Illumina (San Diego, USA) (correction of 40 ntresidues). Sequences were assembled by GATC Biotechusing Newbler, resulting in 50 large contigs, which werearbitrarily connected, providing a sequence of 1000 Kbpapproximately. Lasergene SeqMan Pro V8 (DNAStar,Madison, USA) was used for all subsequent assembliesperformed by the authors. The genome sequences ofstrains 95043 and 97095-Tigray were obtained by Solexasequencing. They were assembled independently, usingthe 9231-Abomsa sequence as reference, and comparedfor the detection of polymorphic sites. Sequences show-ing insufficient cover (i.e.: less than five reads) as well asthose exceeding twice the expected cover were excludedfrom this comparison. For the design of the MLSA sys-tem, several loci < 800 bp located within different con-tigs and showing multiple polymorphisms were selected.Both SNP and indels were considered, with the excep-tion of indels in homopolymer sequences.To identify the nature of the sequences corresponding to

each locus, extended sequences including 2000 flankingnucleotides on either side of each locus were analysedusing Vector NTI Advance™ 11.0 (Invitrogen Corporation,Carlsbad, USA). All identified open reading frames (ORF)were translated using the Mycoplasma/Spiroplasma geneticcode and were compared by protein BLAST with the nonredundant databases through the NCBI server [17].

PCR and sequencingAmplification of each of the locus sequences was per-formed in 50 μL reactions containing: 1 × Taq Buffer(Qiagen) with a final concentration of 1.5 mM MgCl2; 150μM dCTP and dGTP; 300 μM dATP and dTTP; 0.4 μMeach primer, 1 U Taq polymerase (Qiagen) and 1 ng oftemplate. PCR reactions consisted in an initial denatura-tion step of 2 min at 94°C, followed by 35 cycles of 15 s at94°C, 15 s at each corresponding annealing temperatureand 30 s at 72°C. A final extension step was maintainedfor 5 min at 72°C. Primer sequences, annealing tempera-tures and PCR product sizes are shown in Table 2. Thesame primer pairs were used for sequencing of the

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 2 of 10

corresponding PCR products by Beckman Coulter Geno-mics (Takeley, UK). The sequences obtained from eachcorresponding forward and reverse primer were assembledusing Vector NTI Advance™11.0 (Invitrogen Corpora-tion) and the extremities showing single strand sequences,as well as primer or aberrant sequences, were trimmed.All the corrected sequences obtained for each locus werealigned using ClustalW (Vector NTI) and were trimmedto the same size.

Diversity analysisThe locus sequences corresponding to each strain wereconcatenated head-to-tail for diversity analyses con-ducted using Darwin 5.0 [18]. A distance tree was con-structed using the neighbour-joining algorithm. Since

our sample of strains was not a random representationof the Mccp population, the “unweighted” option waschosen. Because sequences were highly similar, theeffect of multiple substitutions was considered negligibleand no correction was applied to dissimilarities. The“pairwise gap block correction” option was selected witha minimal length for gap blocks of 1 nt. This impliedthat all consecutive gaps, starting from one nucleotide,were considered as a single event. Bootstrap analysiswith 1000 replicates was performed.

ResultsChoice of loci for MLSAThe H2 locus, which showed 12 polymorphic sitesand had proven to be a valuable tool for Mccp typing

Table 1 List of Mccp strains analysed and corresponding 16S rDNA, H2 locus and MLSA types

Straina Supplier Countryb Locationb Year Species 16S rDNA H2 MLSA

97095-Tigray NVI-E Ethiopia Tigray (North) 1997 Goat IA1a Aa 1-010

9277-PF1 VRA Sudan NK < 1992 Goat - Aa 1-010

99108-P1 SVS Eritrea Adi-Keshi, Gash-Barka 1999 Goat - Aa 1-010

04012 AWWP Qatar Doha, Al Wabra 2004 Wild Goat - Aa 1-010

M74/93 NVI-S Uganda South East 1993 Sheep IA Aa 1-020

M79/93 NVI-S Uganda East 1993 Goat IA Aa 1-020

8789 LRVZF Chad Karal, Dandi 1987 Goat IB Ca 2-010

94156 LRVZF Chad N’Djamena 1994 Goat - Ca 2-010

05021 VRA Sudan Darfur, Nyala < 2005 NK - Ca 2-010

95043 LABOCEL Niger Goure (East) 1995 Goat I Cb 2-020

M1601 LVRI China Gansu (Centre) 2007 Goat - A 3-010

09018 CIRAD Tajikistan Rogun district 2009 Goat - D 3-020

C550/1 CVRL UAE Dubai 1991 NK IIA D 3-030

Gabes CIRAD Tunisia Gabes (South) 1980 Goat IIB2 B 4-010

LKD CIRAD Tunisia Kebili, Douz (South) 1980 Goat IIB2 B 4-010

Gabes/102p CIRAD Tunisia Gabes (South) 1980 Goat IIB2a (SR) B 4-010

9081-487P MAF-O Oman NK 1990 Goat II B 4-010

07033 FU Turkey Elazig (East) 2007 Goat - B 4-010

7/2 MRI Oman NK 1988 Goat - B 4-020

97097-Errer NVI-E Ethiopia Errer (East) 1997 Goat IIB5 (SR) Ac 5-010

AMRC-C758 AU Sudan NK 1981 Goat IIB A 5-020

Yatta B NVI-S Kenya Eastern, Yatta < 1997 NK IIB4 A 5-020

F38T Type strain Kenya NK 1976 Goat IIB Ab 5-030

94029-C5 AVS Oman NK 1994 Goat - A 5-040

91039-C3 NVI-E Ethiopia NK 1991 Goat - A 5-050

9231-Abomsa CIRAD Ethiopia Godjam (West) 1982 Goat IIB1 A 5-060

92138-CLP1 NVI-E Ethiopia NK 1992 Goat - A 5-060aAll strains are Mccp isolates with the exception of strain 09018, which corresponds to extracted DNA from a CCPP clinical sample, from which Mccp could notbe isolated, and Gabes/102p, which was obtained from strain Gabes after 102 in vitro passages. Strains that were not included in a previous study based on H2locus typing [8] are underlined. Italicised are related isolates or variants used to analyse the stability of the MLSA markers and regarded as a single strain.Therefore, only 25 strains were considered for molecular epidemiology analysis.bThe country and location are those of isolation. However, in two cases, diseased animals were known to be imported from another country: Strain 99108-P1 wasisolated in Eritrea from animals coming from Tigray, north Ethiopia, whereas strain 7/2 was isolated in Oman though actually coming from Turkey [27].

Abbreviations: AU: Aarhus University, Denmark; AVS: Agriculture and Veterinary Services, Oman; AWWP: Al Wabra Wildlife Preservation, Qatar; CIRAD, France;CVRL: Central Veterinary Research Laboratory, United Arab Emirates; FU: Firat University, Turkey; LABOCEL: Laboratoire Central de l’Elevage de Niamey, Niger;LRVZF: Laboratoire de Recherches Vétérinaires et Zootechniques de Farcha, Chad; LVRI: Lanzhou Veterinary Research Institute, China; MAF-O: Ministry ofAgriculture and Fisheries, Oman; MRI: Moredun Research Institute, UK; NVI-E: National Veterinary Institute, Ethiopia; NVI-S: National Veterinary Institute, Sweden;SVS: Senhit Veterinary Service, Eritrea; VRA: Veterinary Research Administration, Sudan; NK: not known; SR: streptomycin resistant mutant.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 3 of 10

[8], was retained for MLSA. The choice of additionalmolecular markers was done according to the analysisof polymorphisms observed among near-completegenome sequences of three Mccp strains correspond-ing to different H2 locus groups: 9231-Abomsa,95043 and 97095-Tigray (Table 1). The aim was tomount up to around 50 polymorphic positions withinseven or eight loci (standard for multi-locus sequencetyping, MLST) to construct a discriminative typingtool.Sequence comparisons between strain 9231-Abomsa

and strains 95043 and 97095-Tigray respectively resultedin detection of over 1000 SNP within a genome ofaround 1000 Kbp. Twenty-two polymorphic loci wereanalysed using three additional strains corresponding tothe most frequently represented H2 locus types: 94029-C5, 07033, and M74/93 (Table 1). This resulted in theselection of seven new variable loci (Table 2).

Organisation of the MLSA lociThe sequences corresponding to the seven new loci wereexamined for functional categorisation by BLAST analy-sis. All except Loc-20 showed the highest similarity tosequences of the Mcc strain California kid (ATCC27343) complete genome (CP000123). Most of the poly-morphic sites were located in intergenic sequences andin what appeared to be pseudogenes, with the exceptionof 3 SNP in Loc-01 and 4 SNP in each Loc-12 and Loc-15, which were located in apparently full CoDing

Sequences (CDS). Loc-01 was found to be homologousto the end of the 1-phosphofructokinase gene (fruK) ofMcc California kid (MCAP_0619) and to the intergenicsequence preceding MCAP_0620. Loc-03 correspondedto the end of a putative lipoprotein gene in Mcc Califor-nia kid (MCAP_0782), which was truncated in Mccp, andto the beginning of a putative PTS system, IIBC compo-nent gene (MCAP_0783), also truncated in Mccp. Loc-11showed greatest similarity to the ornithine carbamoyl-transferase gene (argF) of California kid (MCAP_0654),which was truncated at both N- and C- terminal ends inMccp. Loc-12 corresponded to the N-acetylglucosamine-6-phosphate deacetylase (nagA) of Mcc (MCAP_0438),whereas Loc-15 comprised the spermidine/putrescineABC transporter permease component (potB,MCAP_0201) and permease and substrate-binding com-ponent (potCD, MCAP_0200). Loc-17 matched a putativemembrane protein gene (MCAP_0137) truncated inMccp. For Loc-20, no homologue was found in the Cali-fornia kid genome. The largest ORF in this locus showed34% identity to the Maltodextrin ABC transporter per-mease gene (malC) of Mycoplasma mobile(MMOB3890), though the CDS was extensively truncatedin Mccp. Finally, the organisation of the H2 locus hasbeen previously described [8].

Validation of the stability of the MLSAThe stability of the eight molecular markers selected forMLSA was assessed by analysis of epidemiologically

Table 2 Primer pairs developed in this study and variability of MLSA loci among 25 strains analysed

Locus Primer name Primer sequence (5’-3’) AnnealingT° (°C)

Amplicona

(bp)Locus sequence b

(bp)Variable sites Sequence

types

Loc-01 MLSA-Mccp-01-F GCTTATAGTGTTGTTGATACG 53 694 590 5 5

MLSA-Mccp-01-R GCAATAATCAATTAGCACAG

Loc-03 MLSA-Mccp-03-F ATTCCTCTCATTGAAGTTAC 47 765 664 7 7

MLSA-Mccp-03-R TAGATTAAGAGTCACAATGC

Loc-11 MLSA-Mccp-11-F TGATGGAATTATGTGTAGAGC 53 746 637 6 6

MLSA-Mccp-11-R ATGAACGATCTTGATGTTCC

Loc-12 MLSA-Mccp-12-F GGTATGGAGTTGATTTTGAAAC 58 742 646 4 4

MLSA-Mccp-12-R GCTCCAGCTAAAGCATTATTA

Loc-15 MLSA-Mccp-15-F GGACGAATTTATTTAGTTTCTGCTG 58 802 691 4 4

MLSA-Mccp-15-R ACATTAGTTTGCATACCACCAGTAA

Loc-17 MLSA-Mccp-17-F TAAACCAGAGCAAAACGGTA 58 751 649 8 6

MLSA-Mccp-17-R AACACTAACAATTCCAACAGC

Loc-20 MLSA-Mccp-20-F CTAGTTAATTTTGGAGCCGA 53 781 696 7 7

MLSA-Mccp-20-R CATCAATTGTTGATGAATCG

H2c N/A N/A N/A N/A 2174 12 8

8 concatenatedloci

N/A N/A N/A N/A 6747 53 15

aAmplicon sizes in base pairs correspond to F38T amplificationsbLocus sequence sizes in base pairs correspond to F38T datacFor information regarding H2 locus amplification and sequencing refer to [8]

N/A: does not apply.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 4 of 10

related strains (isolated in nearby locations during anepizootic CCPP episode), as well as a variant obtainedby in vitro passage. Two isolates originating from differ-ent locations but related to the same outbreak in Tuni-sia (Gabes and LKD, Table 1) and a subculture of strainGabes after 102 in vitro passages were analysed for thispurpose. The sequences corresponding to each of theeight loci of these three “variants” were identical, show-ing that the molecular markers were stable and therewere no laboratory-introduced variations.

Molecular typing and geographic distribution of MccpstrainsThe MLSA strategy based on eight loci was extended tothe strains listed in Table 1 with the exception of twostrains used only as controls of the stability of theMLSA markers. The number of variables (SNP andindels) observed within each of the eight loci is indi-cated in Table 2. Fifteen different sequence types (ST)were discriminated among 25 strains, based on 53polymorphisms.The polymorphisms observed within each of the seven

new loci among 25 strains are shown in Table 3. All thesequences were deposited in GenBank (Additional file 1,Table S1). As for the H2 locus, the sequences of eightstrains that had not been previously analysed weredetermined in this study as previously described [8]. Nooriginal sequences were identified and correspondingH2 groups are shown in Table 1.The eight locus sequences corresponding to each

strain were concatenated head-to-tail for sequence dis-tance analysis. A robust tree (showing structured groupssupported by high bootstrap values) was constructedusing the neighbour-joining method (Figure 1). Two dif-ferent lineages and five groups were identified. Lineage Iwas quite homogeneous and comprised two clusters:group 1, including strains from east Africa and an iso-late from Qatar, and group 2, clustering strains fromcentral Africa. Lineage II showed greater heterogeneityand comprised group 3, represented by strains from eastAsia and an isolate from United Arab Emirates, group 4,including strains from north Africa, Turkey and theArabian Peninsula, and group 5, represented by strainsfrom east Africa and an isolate from Oman. A good cor-relation between MLSA groups and geographic originswas observed, with the exception of the Arabian Penin-sula (Qatar, Oman, United Arab Emirates) whereinstrains corresponding to four out of the five differentgroups could be found. Both groups 1 and 5, each cor-responding to a different lineage, were present in eastAfrica. The geographic position of the different groupsand sequence types is displayed in Figure 2.Each of the five groups presented at least two different

ST, allowing a further discrimination into subgroups. In

group 1, the isolates from Uganda could be differen-tiated from all remaining strains. The Ethiopian and Eri-trean strains were actually related since the goatsresponsible for the CCPP outbreak in Eritrea wereimported from the Tigray province of Ethiopia, wherethe Ethiopian strain was isolated (Table 1). Group 2could also be divided into two subgroups. This was theonly clade that was not found in the Arabian Peninsula.Clade 3 was the most diverse, comprising three strainscorresponding to three different ST. The Chinese isolatewas the only one presenting a complete H2 locus,whereas strains from Tajikistan and the United ArabEmirates showed a previously described deletion of theH2 pseudogene [8]. Group 4 was rather homogeneous,comprising only two ST differentiated by a single SNP.One of the Omani isolates was actually of Turkish ori-gin, corresponding to imported goats from Turkey(Table 1). Finally, group 5 was the best represented clus-ter, comprising eight strains corresponding to six differ-ent ST.

DiscussionSequence-based genotyping methods are technicallysimple, objective and portable [19]. The fact that thesemethods do not require isolation of the CCPP agent isparticularly useful considering its fastidious nature.Direct amplification and sequencing from clinical mate-rial is possible even in cases where concomitant bacteriaor antibiotic therapy hamper Mccp isolation, as wasdemonstrated in this work by analysis of a clinical sam-ple from the Tajik outbreak. All these advantages hadalready been taken into consideration by the authorswhen developing an initial tool for Mccp typing basedon the H2 locus [8]. The objective was now to evolvefrom a single locus sequence typing method to the ana-lysis of multiple molecular markers in order to providegreater resolution by considerably increasing variability.Since the limited intraspecies variability of Mccp pre-cluded the use of housekeeping genes, which constitutethe standard molecular markers for MLST, the methodwas adapted to the analysis of other sequences, regard-less of their coding capacity, which is currently knownas MLSA. The selection of new molecular markers forMLSA was done in a comprehensive manner by com-parison of near-complete genome sequences and identi-fication of variable regions distributed along the Mccpgenome. The resulting MLSA scheme based on eightloci revealed the presence of five distinct clades thatwere distributed in two evolutionary lineages.This new MLSA scheme constitutes an important

improvement from H2 locus typing, which was based onthe analysis of a single genomic fragment of over 2000 bp[8]. MLSA resulted in a higher number of polymorph-isms and an increased discriminatory power (53

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 5 of 10

Table 3 Sequence polymorphisms among Mccp strains

Loc-01 Loc-03 Loc-11 Loc-12 Loc-15 Loc-17 Loc-20

Positiona 56 74 258 433 441 54 222 244 528 534-

535

546 579 54 391 566 586 605-

606

625 31 533 542 567 11 406 451 649 18 28 104 157-

158

256 468 529 639 29 112 317 506 575 658 681

F38T G A C A A C G C A - T G C G C G - C A A G G T C T T G C C - A A G T T A G C T C T

97097-

Errer

. . . . . . . . . . . . . . . . . . . . . A . . . . . . . . . . . . . . . . . T .

94029-C5 . . . . . T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . T .

91039-C3 . . . . . T . . . . . . . . . . A . . . . . . . . . . . . . . . . . . . . A . T .

9231-

Abomsa

. . . . . T . . . . . . . . . . A T . . . . . . . . . . . . . . . . . . . A . T .

Gabes . G . . . . . . C . C . . . . . . . . . . A . . . . . . . . . . . . . . . . . T C

09018 . G . . . . . . C . C . . . . . . . . . A A G . . . . . . . G . . G . . . . C T C

7/2 . G . . . . . . C . C . . . . A . . . . . A . . . . . . . . . . . . . . . . . T C

M1601 . G . . . . . T C . C . . . . . . . . . A A G . . . . T . . G . . G . . . . C T C

C550/1 . G . . . . A . C . C . . . . . . . . . A A G . . . . . . . G . . G . . . . C T C

97095-

Tigray

A G G . . . . . C . C T T A . . . . G G A A G T . C A . T . G G A G . G A . C T C

M74/93 A G G . G . . . C . C T T A . . . . G G A A G T . C A . T . G G A G . G A . C T C

95043 A G G T . . . . C T C T T A T . . . G G A A G . C C A . T . G . A G C G A . C T C

8789 A G G T . . . . C T C T T A T . . . G G A A G . C C A . T T G . A G . G A . C T C

aPosition as on F38T sequences.

The polymorphisms found in each of the seven new MLSA loci for each ST are shown, represented by a single strain. Note that strain AMRC-C758 (representing MLSA type 5-020 in Figure 1) is identical to strain F38T by analysis of the sevennew loci shown here, though F38T may be differentiated by H2 locus analysis (MLSA type 5-030). The GenBank accession numbers corresponding to F38T locus sequences are as follows: Loc-01: HQ864744, Loc-03: HQ864761, Loc-11:HQ864776, Loc-12: HQ864786, Loc-15: HQ864807, Loc-17: HQ864814 and Loc-20: HQ864737. All accession numbers may be found in Additional file 1, Table S1.

polymorphisms providing 15 ST) compared to H2 locussequence analysis (12 variables, 8 ST), as calculated bythe comparison of 25 strains analysed (Table 2).Although some correlation was observed, H2 groupswere not always consistent with MLSA groups. As anillustration, H2 locus group A clustered strains belongingto the two distinct MLSA lineages: Aa (correspondingbasically to east African strains from lineage I) and A,

Ab, Ac (mainly representing east African strains fromlineage II). Though allowing a perfect correlationbetween cluster and geographic origin, H2 typing did notreveal the existence of two different lines of descent ineast Africa. In conclusion, MLSA provided higher resolu-tion for molecular typing whilst overcoming the bias ofindividual gene specificities, therefore better representingthe evolution of Mccp strains.Another study on the molecular evolution of Mccp

had been previously performed based on 16S rDNAsequences [7,16]. Fifteen of the strains analysed by 16SrDNA were also analysed in this work, allowing adirect comparison of the results (Table 1). First, thetwo lineages evidenced by MLSA analysis correlatedwell with those obtained using 16S rDNA sequences,supporting the evolutionary significance of these twomajor lines of descent, which were also supported bythe analysis of several housekeeping genes [6]. More-over, the MLSA strategy provided similar resolutionthan that obtained by 16S rDNA analysis, thoughbased on a much higher number of polymorphisms.When taking into consideration the 15 strains com-mon to both studies, 12 ST were discriminated by 16SrDNA analysis according to 16 polymorphisms,whereas 10 ST were obtained by MLSA based on 49polymorphisms. However, this must be regarded withcare as two of the ST discriminated by 16S rDNA typ-ing were directly related to the presence of a singlemutation conferring streptomycin resistance. Since theappearance of this mutation was not related to the nat-ural evolution of the strains baring it but to the selec-tive pressure resulting from exposure to streptomycin,either in the field or in the laboratory, the correspond-ing SNP should not be retained for molecular epide-miology analysis. This would reduce to 10 ST thenumber of evolutionary significant types obtained by16S rDNA. In conclusion, MLSA provided the sameresolution as 16S rDNA analysis for molecular typingof Mccp strains, while allowing the discrimination offive evolutionary groups consistent with CCPP epide-miological data.A discriminative tool for the molecular typing of

Mccp strains has been developed here. The main limita-tion to study the molecular epidemiology of CCPPremains the lack of Mccp strains or DNA samples foranalysis. However, the analysis of genetic data generatedin this work has provided some answers to the initiallyposed questions.A good correlation between MLSA groups and geo-

graphic origins of the strains was observed. The onlyexception to this was the Arabian Peninsula, whereinMccp strains corresponding to several evolutionarygroups were found. This may be explained by the fre-quent importation of animals from diverse origins,

95043-NR

C550/1-AE

09018-TJ

M1601-CH

F38-KE

Gabes-TN

8789-CD

7/2-OM

91039-C3-ET

Abomsa-ET

94029-C5-OM

Tigray-ET

Errer-ET

M74/93-UG

0 0.001

96

88

100

99

99

100

97

99

AMRC-C758-SD

2-020

2-010

1-0101-020

3-0103-020

3-030

5-030

4-010

4-020

5-010

5-020 5-040

5-050

5-060

Lineage I

Lineage II

Group 1

Group 2

Group 3

Group 4

Group 5

95043-NR

C550/1-AE

09018-TJ

M1601-CH

F38-KE

Gabes-TN

8789-CD

7/2-OM

91039-C3-ET

Abomsa-ET

94029-C5-OM

Tigray-ET

Errer-ET

M74/93-UG

0 0.001

96

88

100

99

99

100

97

99

AMRC-C758-SD

2-020

2-010

1-0101-020

3-0103-020

3-030

5-030

4-010

4-020

5-010

5-020 5-040

5-050

5-060

Lineage I

Lineage II

Group 1

Group 2

Group 3

Group 4

Group 5

Figure 1 Tree derived from distance analysis of the eightconcatenated MLSA loci. The tree was constructed using theunweighted neighbour-joining algorithm (Darwin 5.0). A singlestrain representing each of the 15 MLSA ST is displayed (see Table 1for strain details). The two sequences presenting a large 960 ntdeletion (09018 and C550/1) were grafted at their respectivepositions after tree construction (discontinuous lines) in order toavoid their influence during tree inference. The root of the tree isrepresented as a bold dot. Bootstrap percentage values werecalculated from 1000 resamplings and values over 80% aredisplayed. The scale bar shows the equivalent distance to 1substitution per 1000 nucleotide positions. The ST were numberedaccording to the group in which they clustered, followed by a threedigit code that should allow intercalating additional ST as newsequences are obtained. Note that the two letter code followingthe strain name represents the corresponding country of isolation:AE, United Arab Emirates; CD, Chad; CH, China; ET, Ethiopia; NR,Niger; OM, Oman; SD, Sudan; TJ, Tajikistan; TN, Tunisia; UG, Uganda.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 7 of 10

particularly for the Muslim feasts celebrated every year.Otherwise, the geographic distribution of the MLSAgroups was quite explicit.A distinct Asian clade was identified by MLSA, repre-

sented by two strains from Tajikistan and China andalso comprising a strain from Dubai. In spite of thesampling limitation, the existence of this clade stronglysuggests that these strains have evolved locally and,therefore, that they have not been introduced recentlyin this continent. CCPP was suspected long ago in con-tinental Asia based on historical clinical descriptions,with substantiating data presented in India already in1914 [20]. The recent declaration of the disease in Taji-kistan should encourage neighbouring countries tosearch for Mccp, enabling a better assessment of thedistribution of CCPP in Asia.A local evolution of strains was also demonstrated in

central Africa, where a single MLSA group wasobserved. Furthermore, this group was restricted to cen-tral Africa, constituting the only clade that was notidentified in the Arabian Peninsula. Although, arguably,this could be attributed to insufficient sampling, the lim-ited animal movements in this region, where transhu-mance is oriented north-south, may well explain the

exclusive presence of indigenous strains. This leads usto reject the assumption that CCPP was introducedfrom east Africa, as it was proposed in 1987 when thedisease was first discovered in Chad [21]. Also, if weconsider the distribution of other contagious diseases ofgoats such as “peste des petits ruminants”, we may sus-pect that CCPP is also present in west Africa. An activesearch for the etiologic agent should be encouraged toelucidate the western limits of the distribution of CCPPin Africa.Isolates from north Africa and Turkey corresponded

to the same MLSA group, which reflects the importanceof Mediterranean trading routes, particularly the expor-tation of animals from Turkey to north Africa and theArabian Peninsula. CCPP has been known for manyyears in Turkey, where it appears to be widespread.Moreover, uncontrolled animal movements in thisregion should raise suspicions regarding the presence ofthe disease in neighbouring countries [22]. Furtherstrains should be typed to assess the variability existingwithin Turkey, while efforts should be made at a regio-nal level to better understand the distribution of the dis-ease in this area. This applies also to north Africa,where the presence of the disease was confirmed in

1-010

1-020

2-0102-020

3-030

3-020

5-010

5-020

5-030

5-040

5-050

2-010

1-010

1-010

4-010

4-010

4-0205-020

5-060

3-010

MLSA group 1

MLSA group 2

MLSA group 3

MLSA group 4

MLSA group 5

Lineage I

Lineage II

4-010

1-010

1-020

2-0102-020

3-030

3-020

5-010

5-020

5-030

5-040

5-050

2-010

1-010

1-010

4-010

4-010

4-0205-020

5-060

3-010

MLSA group 1

MLSA group 2

MLSA group 3

MLSA group 4

MLSA group 5

Lineage I

Lineage II

MLSA group 1

MLSA group 2

MLSA group 3

MLSA group 4

MLSA group 5

Lineage I

Lineage II

MLSA group 1

MLSA group 2

MLSA group 3

MLSA group 4

MLSA group 5

Lineage I

Lineage II

4-010

Figure 2 Geographic origins of the strains analysed in this study. Each strain is represented by a symbol corresponding to its MLSA groupand by its ST. Note that when the exact location was not known the symbols were barred and were positioned arbitrarily in the correspondingcountry. The arrowed lines represent main routes of trade and other animal movements in the different areas.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 8 of 10

1980 in Tunisia [23], though no further studies havebeen published since then.In east Africa two MLSA groups, each belonging to a

different lineage, were identified. Strains belonging toeach of the two evolutionary lines have been spreadingin this region over the last decades and the disease hasrecently reached the Indian Ocean [10].Although the recent confirmation of the presence of

CCPP in continental Asia has provided a better estimationof the distribution of CCPP world-wide, some questionsstill remain. CCPP has only been reported in sixteen coun-tries, while, if we take into consideration reports of clinicaldisease, over forty countries of Africa and Asia may beaffected. The boundaries of the disease in Asia, as well astowards the west and south of the African continent arestill uncertain but, taking into consideration the conta-giousness of the disease and the movements of nomadicgoat herds, CCPP is probably present in central andnorth-east Africa, the Middle East and all the way throughto China. Figure 3 shows an updated map presenting theprobable distribution of CCPP.

Mccp has been isolated from sheep showing clinicaldisease that had been in contact with CCPP-infectedgoats [22,24] and the existence of the disease in wildlifespecies was also demonstrated recently in Qatar [14].Although the origin of this outbreak was not elucidated,it was proposed that the wild species kept in a conserva-tion area may have been contaminated by domesticgoats. Actually, two of the strains here analysed corre-sponded to these reports: a sheep isolate from Uganda[24] and another one originating from a wild goat inQatar [14]. Both of them shared ST with other goat iso-lates, suggesting that the same strains can affect a widerange of species. However, further studies are requiredto identify the genetic determinants of species-specifi-city. Also, the role of these species in the epidemiologyof CCPP is yet to be elucidated.A discriminative MLSA scheme has been designed as

a tool for the molecular epidemiology of CCPP. Itwould be advantageous to form a publicly accessibledatabase that will be enriched by sequences obtained bydifferent laboratories in affected countries. Such a

??

?

?

??

?

?

Figure 3 Probable distribution of CCPP. The countries in which the disease has been described, those in which the etiologic agent has beendetected using molecular tests and those in which it has been isolated are indicated. The arrow indicates the presence of the disease inMauritius, where Mccp was isolated in 2009.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 9 of 10

database could be linked to the websites of the OIE [25]and FAO [26], which provide updated informationregarding new Mccp outbreaks. This new typing toolmay help improve the surveillance and control of thedisease, as well as to trace new epidemics.

Additional material

Additional file 1: Table S1. GenBank accession numbers of locussequences obtained in this study. Displayed are GenBank accessionnumbers corresponding to the sequences of 14 strains representing the14 ST discriminated based on the seven new MLSA loci.

AcknowledgementsThe authors wish to thank Burhan Cetinkaya and his team at ElazigUniversity for strains and continuous collaboration, Hélène Guis for thedistribution map of CCPP and Victoria Chalker for revision of the Englishlanguage.

Author details1CIRAD, UMR CMAEE, F-34398 Montpellier, France. 2State Key Laboratory ofVeterinary Etiological Biology, Lanzhou Veterinary Research Institute, CAAS,Xujiaping 1, Lanzhou 730046, China.

Authors’ contributionsFT and LMS conceived the study and participated in its design andcoordination. YC provided material and data from Chinese isolates. VD andLMS carried out the analysis. LMS drafted the manuscript, with thecollaboration of FT, VD and YC. All authors read and approved the finalmanuscript.

Competing interestsThe authors declare that they have no competing interests.

Received: 5 January 2011 Accepted: 14 July 2011Published: 14 July 2011

References1. Thomas P: Rapport médical sur le Bou Frida. In Publication du

gouvernement général civil de l’Algérie. Edited by: Jourdan A. Algiers; 1873:.2. MacOwan KJ, Minette JE: A mycoplasma from acute contagious caprine

pleuropneumonia in Kenya. Trop Anim Health Prod 1976, 8:91-95.3. Abu-Groun EA, Taylor RR, Varsani H, Wadher BJ, Leach RH, Miles RJ:

Biochemical diversity within the “Mycoplasma mycoides” cluster.Microbiology 1994, 140(Pt 8):2033-2042.

4. Bonnet F, Saillard C, Bove JM, Leach RH, Rose DL, Cottew GS, Tully JG: DNArelatedness between field isolates of Mycoplasma F38 group, the agentof contagious caprine pleuropneumonia, and strains of Mycoplasmacapricolum. Int J Syst Bacteriol 1993, 43:597-602.

5. Costas M, Leach RH, Mitchelmore DL: Numerical analysis of PAGE proteinpatterns and the taxonomic relationships within the “Mycoplasmamycoides cluster”. J Gen Microbiol 1987, 133:3319-3329.

6. Manso-Silvan L, Perrier X, Thiaucourt F: Phylogeny of the Mycoplasmamycoides cluster based on analysis of five conserved protein-codingsequences and possible implications for the taxonomy of the group. IntJ Syst Evol Microbiol 2007, 57:2247-2258.

7. Pettersson B, Bolske G, Thiaucourt F, Uhlen M, Johansson KE: Molecularevolution of Mycoplasma capricolum subsp. capripneumoniae strains,based on polymorphisms in the 16S rRNA genes. J Bacteriol 1998,180:2350-2358.

8. Lorenzon S, Wesonga H, Ygesu L, Tekleghiorgis T, Maikano Y, Angaya M,Hendrikx P, Thiaucourt F: Genetic evolution of Mycoplasma capricolumsubsp. capripneumoniae strains and molecular epidemiology ofcontagious caprine pleuropneumonia by sequencing of locus H2. VetMicrobiol 2002, 85:111-123.

9. Ozdemir U, Ozdemir E, March JB, Churchward C, Nicholas RA: Contagiouscaprine pleuropneumonia in the Thrace region of Turkey. Vet Rec 2005,156:286-287.

10. Srivastava AK, Meenowa D, Barden G, Salguero FJ, Churchward C,Nicholas RA: Contagious caprine pleuropneumonia in Mauritius. Vet Rec2010, 167:304-305.

11. Awan MA, Abbas F, Yasinzai M, Nicholas RA, Babar S, Ayling RD,Attique MA, Ahmed Z, Wadood A, Khan FA: First report on the molecularprevalence of Mycoplasma capricolum subspecies capripneumoniae(Mccp) in goats the cause of contagious caprine pleuropneumonia(CCPP) in Balochistan province of Pakistan. Mol Biol Rep 2010,37:3401-3406.

12. Amirbekov M, Murvatulloev S, Ferrari F: Contagious caprinepleuropneumonia detected for the first time in Tajikistan. EMPRESTransboundary Animal Diseases Bulletin, FAO Animal Production and HealthDivision 2010, 35:20-22.

13. Li Y, Zhang JH, Hu SP, Wang L, Xin JQ: Reclassification of the four Chinaisolated strains of the pathogen for contagious caprinepleuropneumonia. Wei Sheng Wu Xue Bao 2007, 47:769-773, (in Chinese).

14. Arif A, Schulz J, Thiaucourt F, Taha A, Hammer S: Contagious caprinepleuropneumonia outbreak in captive wild ungulates at Al WabraWildlife Preservation, State of Qatar. J Zoo Wildl Med 2007, 38:93-96.

15. Lorenzon S, Manso-Silvan L, Thiaucourt F: Specific real-time PCR assays forthe detection and quantification of Mycoplasma mycoides subsp.mycoides SC and Mycoplasma capricolum subsp. capripneumoniae. MolCell Probes 2008, 22:324-328.

16. Heldtander M, Wesonga H, Bolske G, Pettersson B, Johansson KE: Geneticdiversity and evolution of Mycoplasma capricolum subsp.capripneumoniae strains from eastern Africa assessed by 16S rDNAsequence analysis. Vet Microbiol 2001, 78:13-28.

17. BLAST: Basic Local Alignment Search Tool. [http://blast.ncbi.nlm.nih.gov/Blast.cgi].

18. DARwin Dissimilarity Analysis and Representation for Windows. [http://darwin.cirad.fr/darwin/Home.php].

19. van Belkum A, Tassios PT, Dijkshoorn L, Haeggman S, Cookson B, Fry NK,Fussing V, Green J, Feil E, Gerner-Smidt P, Brisse S, Struelens M, EuropeanSociety of Clinical Microbiology and Infectious Diseases (ESCMID) StudyGroup on Epidemiological Markers (ESGEM): Guidelines for the validationand application of typing methods for use in bacterial epidemiology.Clin Microbiol Infect 2007, 13(Suppl 3):1-46.

20. Walker GK: Pleuro-pneumonia of goats in the Kangra district, Punjab,India. J Comp Pathol 1914, 27:68-71.

21. Lefevre PC, Breard A, Alfarouk I, Buron S: Mycoplasma species F 38 isolatedin Chad. Vet Rec 1987, 121:575-576.

22. Cetinkaya B, Kalin R, Karahan M, Atil E, Manso-Silvan L, Thiaucourt F:Detection of contagious caprine pleuropneumonia in East Turkey. Rev SciTech 2009, 28:1037-1044.

23. Perreau P, Breard A, Le Goff C: Experimental infection in goats caused bymycoplasma strain F.38 (contagious caprine pleuropneumonia). AnnMicrobiol (Paris) 1984, 135A:119-124, (in French).

24. Bolske G, Johansson KE, Heinonen R, Panvuga PA, Twinamasiko E:Contagious caprine pleuropneumonia in Uganda and isolation ofMycoplasma capricolum subspecies capripneumoniae from goats andsheep. Vet Rec 1995, 137:594.

25. OIE, World Organisation for Animal Health. [http://www.oie.int/en/].26. FAO, Food and Agriculture Organization of the United Nations. [http://

www.fao.org/].27. Jones GE, Wood AR: Microbiological and serological studies on caprine

pneumonias in Oman. Res Vet Sci 1988, 44:125-131.

doi:10.1186/1297-9716-42-86Cite this article as: Manso-Silván et al.: Multi-locus sequence analysis ofMycoplasma capricolum subsp. capripneumoniae for the molecularepidemiology of contagious caprine pleuropneumonia. VeterinaryResearch 2011 42:86.

Manso-Silván et al. Veterinary Research 2011, 42:86http://www.veterinaryresearch.org/content/42/1/86

Page 10 of 10